Enhancing Transparency: AI Applications for Detecting Cheating and Predicting Player Attrition in Online Gaming

The online gaming industry, a thriving multibillion-dollar sector with a vast global audience, has greatly benefited from the integration of AI technology. However, the widespread adoption of black-box AI models in gaming systems has raised concerns regarding their lack of transparency and accountability. In this article, we address this issue by introducing Explainable AI (XAI) concepts tailored to online gaming environments. We propose a GXAI (Game Explainable AI) workflow that combines the rich insights from multiview data sources with the transparent nature of multiview black-box models. Our approach encompasses four specific classifiers and explainers, each offering unique perspectives: character portrait view, behavior sequence view, client image view, and social graph view. Through experiments conducted on real-world datasets, focusing on cheating detection and player churn prediction, we demonstrate the accuracy of our classification methods and the coherence of our explanations.